Prediction of Coronary Plaque Progression Using Data Mining and Artificial Neural Networks

Coronary artery disease represents one of the most significant health burdens worldwide. As the onset of the disease is multifactorial in nature, physicians are often struggling with determining the rate of progression of the arterial narrowing caused by buildup of plaque. Computational models have brought upon a significant shift in the paradigm and the advent of Big Data and machine learning has enabled far better understanding of disease dynamics. This study is based on a cohort of patients recruited through SMARTool project for whom an extensive monitoring system was set up. In order to select for the most influential parameters on the progression of coronary atherosclerosis, different feature selection algorithms were used. The dataset used for development of the system for prediction of coronary plaque progression consisted of demographic data, data considering comorbidities and different blood cholesterol parameters. The developed artificial neural network showed significant strength in diagnosing progression of coronary arterial plaque, and features selected within this study indicate the high potential of machine learning to be used in clinical practice as well as that specific types of cholesterol are important markers impacting plaque progression.